The monthly ICM poll for the Guardian is out tonight and has topline figures of CON 34%(+2), LAB 38%(+2), LDEM 12%(-2), UKIP 8%(-1). Changes are since ICM’s last poll, conducted at the beginning of the Lib Dem conference. The Labour lead is unchanged, with the Conservatives and Labour both up slightly, but not significantly, following their conferences.

The rest of the poll included questions about Ed Miliband’s price freeze for energy bills (61% support the promise, 30% are opposed), Royal Mail privatisation (29% support, 63% oppose) and newspapers publishing details of security service surveillance (58% think papers should back off, 34% think they have a duty to report it. Despite what the Guardian report says about a contrast, there isn’t actually a vast difference from what YouGov found in the Sunday Times – they did find more people opposed to the security services monitoring electronic communications than supporting it, but by 43% to 35% people thought the leaking of information about surveillance was a bad thing).

UPDATE: Tom Newton Dunn at the Sun has also tweeted out tonight’s YouGov figures, which are CON 37%, LAB 38%, LD 10%, UKIP 10%. The Labour lead of just one point is the smallest YouGov have shown for a couple of weeks, but usual caveats apply – sure, it could be the first sign of a narrowing lead, or it could just be normal variation within the margin of error.

103 Responses to “ICM/Guardian – CON 34, LAB 38, LD 12, UKIP 8”

1 2 3
  1. It’s on a knife edge…

  2. Other polls roughly agree with L 38, C 34, but ICM for the Guardian has LibDem 50% more than UKIP on only 8? Leaving aside the actual figures, the equivalent would be a poll showing something like Conservative 42, Labour 28, and who would believe that?

  3. NickP
    What is (on a knife edge)?

    Those who want to I suppose.

  4. and newspapers publishing details of security service surveillance (58% think papers should back off, 34% think they have a duty to report it.

    So, polling strongly opposes newspapers publishing details of state security arrangements. Not surprised at this, as it could still backfire at some stage and make us less safe.

  5. Can anyone remind me what the MOE is? Is it about 3%, and the same for all values?

    What I’m getting at is that if a party on 35% has an MOE of 3%, then their ‘true’ VI could be anywhere from 32 to 38%, but a party on 10% could have a ‘true’ VI of anywhere from 7-13%, which is nearly a 100% variation of their own vote if you see what i mean.

    So though the UKIP VI looks low on 8% compared to other polls, it may not mean anything at all.

  6. Rich

    Just because people believe it doesn’t make them correct – you seem to correlate the ‘making us less safe’ with polling results!

    The UK population have shown over the last 30 years that we are happy to believe all that the security services say, bit sad really

  7. Pete B

    Can anyone remind me what the MOE is? Is it about 3%, and the same for all values?

    We can’t be sure till we see the tables which won’t be till tomorrow at the earliest (later if there are more questions that they are going to use as the basis of news stories). But in September the pre-adjustment figures were based on a sample size of only 460. Although ICM usually ask around a thousand people by phone, many of these will say they won’t vote (11%)[1] don’t know how they will (23%) or refused to say for whom (10%). Many of the rest who give a Party will have their vote downgraded if they say they are not sure they will vote or if they didn’t vote in 2010.

    Reducing the effective sample size to 460 means that MoE is increased from around 3 which is what it would be if all 1000 gave an answer to around 4.5. However this is the MoE for 35-ish percentages (Lab and Con). For 10-ish percents (Lib Dem, UKIP) the MoE will be around 3.

    To make an estimate of how people will actually vote in the next GE, ICM then do some reallocation of the Don’t Knows and Refused according to their 2010 vote (if they voted and told ICM how) but this does not affect the MoE.

    [1] September percentages, but will give an idea.

  8. @Pete B

    I think ICM weight by past vote. Tends to hurt UKIP, and probably flatter the,Tories a,little (at UKIP’s expense).

  9. Rich
    “So, polling strongly opposes newspapers publishing details of state security arrangements. Not surprised at this, as it could still backfire at some stage and make us less safe.”

    What could backfire? The polling result or what you really were trying to say?

  10. but do the 58% feel strongly? Do the 34% feel strongly, only those that feel strongly will make a difference to VI

  11. Rich

    Funny how those newspapers so jealously protective of their own independence and how significant this is to hold government to account, are the self same newspapers that are so strongly against holding big government surveillance and security to account.

  12. @ Howard (From previous thread)

    The Greeks were not the only one… Italy and Belgium also had interesting ways of meeting the Maastricht Treaty criteria…

    The UK actually met them for quite a long time (well, interest rates were a bit dodgy). The key thing in UK public finance accounting is the arbitrary classification of investment and current expenditure. Both sides have used it for political games (and vote buying or to put it differently: influencing VI).

    The French have so many extra budgetary state funds in public finance that make the IMF weep when publishing the stats (perhaps CL can help them though now).

  13. @Pete B

    The confidence interval for a proportion is as follows:

    z x SQRT (p x (1-p)/sample size)

    z = 1.96 for a 95% confidence interval (that is based on a normal distribution curve 95% of values fall within +/- 1.96 standard deviations.

    p = proportion (for Labour @ 40% p = 0.40)

    So if a poll of 1000 people ends with 40 % saying they will vote Labour there is a 95% likelihood the true proportion is:

    = 1.96 x SQRT (0.40 x (1 – 0.40) /1000)

    = 1.96 x SQRT (0.40 x 0.60/1000)

    = 1.96 x SQRT (0.24/1000)

    = 1.96 x sqrt (0.00024)

    = 1.96 x 0.0155 (4 dp)

    = 0.0304 (4 dp)

    That means there is a 95% likelihood that the true Labour VI is 40% +/- 3.04%

    The Lib dems on 10 % are:

    = 1.96 x sqrt(0.10 x (1 – 0.10)/1000)

    = 1.96 x sqrt (0.10 x 0.90/1000)

    = 1.96 x sqrt(0.00009)

    = 1.96 x 0.0095 (4 dp)

    = 0.0185 (4 dp)

    Therefore there is a 95% chance that the LD VI is at 10% +/- 1.85%

  14. ICM Research interviewed a random sample of 1,004 adults aged 18+ by telephone on 11-13th October 2013. Interviews were conducted across the country and the results have been weighted to the profile of all adults.

  15. @Pete B

    The true MOE for a samples of 1000, from 0.1 to 0.9 is as follows:

    Proportion – 95% CI (+/-)
    0.1 – 0.0186
    0.2 – 0.0248
    0.3 – 0.0284
    0.4 – 0.0304
    0.5 – 0.0310
    0.6 – 0.0304
    0.7 – 0.0284
    0.8 – 0.0248
    0.9 – 0.0186

    With statistics you can really can disappear up your own backside just to get the most accurate number.

    However, often things are simplified to to make it much easier, while remaining close enough to the true figure to be worth doing.

    Also, many aspects of polling are not totally reliable, and the whole thing is based on common assumptions that might not be entirely true. Therefore getting obsessed with an MOE of +/- 1.86 vs an MOE of 3.11 isn’t worth it.

  16. RogerMexico and Catmanjeff
    Thanks for the clarification. I had an idea the MOE would be lower for smaller VI, but couldn’t remember how to prove it.

  17. Laszlo

    Thanks. I am aware, for instance, that the rolling performance of the NL deficit (to August) had it down to 2.5% of GDP but that was significantly influenced by the sale of mobile phone frequencies.

    Is that Capital or Revenue? I mean, you are essentially selling something you did not have or were unaware that you did have something and suddenly discovered that you could sell it?.

    One could easily argue that the fresh air in some countries (Scotland?) could be very profitable.

  18. If the security poll fit the narrative, you guys would be jumping all over it. Because it shows people are against leaking of state security arrangements, you just ignore the poll under the rather sanctimonious liberal left we know best. You cant have it both ways, it is a polling site after all.

  19. @Rich

    “But, it may be exaggerated, or it may not be, but there are a volume of people who are happy not to work and spend a life on benefits. I have seen evidence of this myself.

    What doesn’t help the argument is some on the liberal left who deny the problem at all and say its a Tory masterplan to turn people against each other. It’s ridiculous, and no better than people on the right who exaggerate the problem. It’s going to be incredibly difficult and complex to ever crack this problem now given the changing world, globalisation, population etc. (just look at youth employment in parts of Europe”


    Thing is Rich, you can focus on those who game the system, gaming which is a pretty inevitable outcome amongst some when facing a hopeless situation – but the problem is there aren’t the jobs anyway. It’s not going to solve the problem of big welfare bills. Decent-paying jobs solves the problem – we didn’t have big welfare issues when we had full employment – and as you point out, creating those is tricky, especially once you’ve let a region decline, which is one reason why it’s a good idea to try to preserve them through a downturn in the first place, like some of our rivals do.

    But a lot of this is political. Building affordable housing could be very cost-effective but politically unpalatable. Equally, why focus on those struggling on benefits who’re gaming the system because in a position of some hopelessness, when at the top there’s tax avoidance, MPs’ expenses, and bankers getting bonuses while taking down the economy, causing unemployment and upping the welfare bill?

    Also, a lot of these welfare issues arose in the eighties, so it’s hardly a leftie plot. I was reading the New Scientist today, and it was talking about another worrying reason to be concerned about unemployment…

  20. Tweet from:
    Tom Newton Dunn [email protected] 13m

    YouGov/Sun poll tonight: Labour lead down to 1% again. LAB 38%, CON 37% LD 10%, UKIP 10%. An outlier, but suggests contraction under way.

  21. @carfrew,
    i agree in the most part. But the social problems have gone so deep rooted i wonder how we can turn it round. I don’t always buy the 80s fault, as so many other countries who didn’t have Thatcherite policies are suffering very similar dreadful social and youth unemployment issues. I couldn’t agree more on the housing front though.

  22. And the Sun tweets a “gud ‘un” again…

    Is that the highest Tory score we’ve had all year?

    It is starting to seem like the public aren’t really paying that much attention to the politicos. Or at least that the plus points of each side are cancelling each other out.

    Perhaps it’s true that it’s really “the economy, stupid”.

  23. Evening everyone

    Is that tweet for real?

  24. @Rich

    I think we’d agree on a fair bit, the main sticking point is over whether targeting the gaming isn’t tackling symptoms rather than causes. Even without gaming and social problems, there still aren’t the jobs, hence, big welfare bill regardless, made bigger during banking crisis and then the cuts.

    Thus, a focus on gaming, as opposed to jobs, is unlikely to lead to a solution to the welfare bill.

    (With the eighties thing, I was thinking in particular about the mass migration onto incapacity that we hear about, though I’m not sure how big a phenomenon it was, and how much it was encouraged to keep headline figures down…)

  25. @ Neil A

    37% for the Tories would their highest polling since March 2012.

    I think the YG polling data has been showing an increasing Tory sample for a number of weeks. Then through weighting this has been reduced.

    The economy is improving and the government are receiving more positive media coverage. Tories attacking illegal immigrants is popular with some, particularly those who may have looked at UKIP.

  26. Howard

    Taking back control of the money supply would reduce the deficit very quickly(assuming you want to reduce it) if the govt had keep full control over the money supply it would be 2.1 trillion pounds better off over the course of the last 40 years, that’s equivalent to three years of absolutely no taxes

  27. @NEIL A

    “Perhaps it’s true that it’s really “the economy, stupid”.”


    Well, I’ve made the case before, that for us in the UK, services can matter rather a lot too, but I posted a poll the other day showing public satisfaction with a number of services seems to have grown since the cuts began, which is quite a win for Tories. Question is whether councils etc. can maintain that with further cuts…

    (people weren’t very happy about the potholes thing however, as unemployment might expect…)

  28. What would the seat count be on those figures?
    FPTP is brutal ain’t it.

  29. As ONE might expect…

  30. @RogerMexico and Catmanjeff and RAF
    Thanks for the comments.
    There are two ways in which a measurement can differ from the “true” value.
    The first is random error, which in polling is largely dependent on sample size.
    The second is that the measurement may be subject to one or more systematic errors. One such will always shift the measurement in the same direction, like measuring a length with a stretched tape, or timing a 100m race starting the watch on the sound of the gun rather than on the puff of smoke. If there is more than one source of systematic error, some may act in one direction, others in the opposite direction. This may give an impression of the result being subject to larger random errors than sample size might indicate, but this is somewhat illusory, for if one systematic error is identified and removed, all the results are shifted in the opposite direction.
    However, all the polls purport to be trying to measure VI. The range of results shows clearly that systematic errors are more important than random errors in accounting for differences between pollsters, but there does not seem to me to be enough effort put into identifying them.
    eg as RM posted, “ICM then do some reallocation of the Don’t Knows and Refused according to their 2010 vote” and RAF said “I think ICM weight by past vote. Tends to hurt UKIP, and probably flatter the,Tories a,little ” and I suppose other pollsters apply there own “corrections”. But fudging the figures in the light of past experience as to what gets nearest to the eventual actual result is not the same as knowing which part of polling technique throws the answers high and which throws them down.

  31. @Rich
    “if the security poll fit the narrative, you guys would be jumping all over it. Because it shows people are against leaking of state security arrangements, you just ignore the poll under the rather sanctimonious liberal left we know best. You cant have it both ways, it is a polling site after all.”

    Not really. I would expect people to think that way. It’s entirely.unsurprising. My point was that newspapers say they are there to hold the state to account but only do so when it suits their politics. And i’m a liberal not a leftie. Given this is a country built on the fight over centuries for individual liberty from the state, I find it somewhat disappointing that people are so sanguine about returning those freedoms.

  32. That YouGov poll (37, 38, 10, 10) is more a sign of Tory strength than Labour weakness.

    The last 1% poll lead back on 19th Sept had Con 34, Lab 35, LD 11, UKIP 11. By contrast, in this one, Lab still hasn’t dipped below its post conference floor, which had been reestablished at 38. Note also what’s happening to the “Others” in the meantime (SNP, PC, Green, etc). These %s imply that they’re down to 5%, compared to just 9% collectively in the previous 1% poll.

    That said, I agree fully with @DrunkenScouser (FPT) that Rachel Reeves faffing around pretending that she’s IDS on steroids is a lousy way to promote Labour’s jobs guarantee for the long term unemployed.

  33. “…Rachel Reeves faffing around pretending that she’s IDS on steroids…”

    I blame Ian Katz.

    Next YouGovPrediction:

    Con 34.0%
    Lab 38.9%
    LD 9.7%

  34. Recovery – or no recovery. Main thing is the Clangers are back !
    The soup dragon will be included. What more could I wish for ? Made my day, though my grandkids will think I am batty for being a fan. :-)

  35. Outlier or not, a 1 % is truly worrying for Labour. How long until we see the Tories ahead, and then even maintain a lead?

    You’d have to suggest that if in 12 months time the polls are level, the Conservatives would be far happier.

  36. @Ozwald

    Good news. A little before my time, but I did see it in 80s. And it was genius. Although, in fairness I was more part of the DangerMouse generation.

  37. @Jack R

    The Tories can’t sustain 35+ without a UKIP dip.

    “But a lot of this is political. Building affordable housing could be very cost-effective but politically unpalatable.”

    I wonder if EM gives a tinker’s about any unpalatability of increased building of affordable housing, given the figure of 200,000 p.a. which he has provided as policy. Perhaps you need to add, the “new towns” developments and expansion of existing towns, urban development in decayed areas of London and other cities and related infrastructure, as primary and second phase employment creators. And then, the taking up of unemployment from below, in training and apprenticeships.

  39. May I repeat my question from the last thread?
    What is regarded as an acceptable level of unemployment – I mean from a welfare standpoint, rather than as a lever? Is 4 to 5% in the active work force, and 7-8% among young entries to the work force, a level which would represent an acceptable margin of people between jobs and between school and jobs?

  40. @John P

    Well, you may think Miliband cares not for palatability, if you think he’s never dodged a question, doesn’t consider electoral impact of policies, doesn’t use focus groups etc.

    And I don’t need to add anything. Many promises have been made and broken, by Labour as well. And housebuilding does have a politically unpalatable side, since it may not happinate boomers.

    That said, politically, as someone else pointed out at the time, Miliband could use housebuilding while mitigating VI effects, by building a shed load in Tory heartlands…

  41. We have a non-polldrums poll for once.


    You can’t beat Jamie and the Magic Torch.

  42. My prediction is we will see Labour back down to 35’s over the next 2 weeks. Their recent bounce was a ‘core vote’ strategy during the conference and now Yvette Cooper with her ‘sensible’ verdict on the Tory immigration bill, and Rachel Reeves’ “tougher than the Tories on benefits” just killed that bounce. We will see Labour voters moving back to the undecided column. When will we see a UKIP of the left?

    With the economy improving, I can’t see any way but up, or at least stability for the Tories, so I predict a number of polls showing Tory leads sometime over the next few weeks. (Assuming the US don’t destroy the world economy in the meantime)

  43. PETE B That’s not how binomial statistics work. MOE is smaller for smaller parties. Roughly speaking if N individual people in a poll say they will vote for UKIP then this comes with a margin of error of +/- sqrt(N)

    Answer the question, matey.

  45. “Evening everyone

    Is that tweet for real?

    Ah…. like the first cuckoo, a sine of a tight poll.

    Richard: bet you are wring.

  46. Eh? What question John? You didn’t ask me a question…

  47. @John P

    Oh, do you mean the question in the last thread you addressed to Rich, lol. There’s a degree of unemployment that is somewhat inevitable – churn, people unable to work, people choosing to take time out, maybe for family etc. – though offhand I’m not sure what that figure might currently be.

    Beyond that, needless involuntary employment would be something that has commonly understood drawbacks so one might wish to bear down on it. Unless one starts down the citizen income road of course, which is a whole other kettle of fish…

  48. @ John P

    There is also the question of things like what to do about carers etc., so it’s hard to put a figure on the percentage of natural/inevitable unemployment. But still, one would seek to have enough jobs for all after taking into account churn etc.

    Thanks. I think it is the figure, which you’re right needs to be verifiable, for the unemployed, 4% good, 6% well Ok, 8% bad that signifies something in the distributive economy, and VI.

  50. The last 1% lead for Labour was 17 polls ago folks.

    Do bear in mind that about 1 in 20 (a result of a 95 % confidence interval) polls will look a bit off.

    The fact that a two odd looking polls (odd against recent data) are 17 polls apart looks quite normal.

1 2 3